personal memory agent
Fix insight model display to match actual context used
The display was using 'insight.{key}' but send_insight uses
'insight.{key}.{output_type}', causing tier pattern mismatch.
For entities, this showed gemini-3-flash-preview instead of
the actual gemini-2.5-flash-lite being used.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
+3
-2
+3
-2
think/insight.py
+3
-2
think/insight.py
···
435
435
436
436
prompt = insight_prompt.text
437
437
438
-
# Resolve provider for display (routing happens inside send_insight)
438
+
# Resolve provider for display (must match context used in send_insight)
439
439
from muse.models import resolve_provider
440
440
441
-
_, model = resolve_provider(f"insight.{insight_key}")
441
+
display_output_type = "json" if output_format == "json" else "markdown"
442
+
_, model = resolve_provider(f"insight.{insight_key}.{display_output_type}")
442
443
day = args.day
443
444
size_kb = len(markdown.encode("utf-8")) / 1024
444
445